Theiling Online    Sitemap    Conlang Mailing List HQ   

CHAT: THEORY/CHAT: Re: Jackendoff's "Semantic (?) Structures"

From:And Rosta <a.rosta@...>
Date:Wednesday, April 21, 2004, 23:21
Philippe Caquant:
> At least Father Amazon brought me Ray Jackendoff's > "Semantic structures", and I opened it with much > curiosity. I for sure haven't read it thoroughly yet, > and even less thoroughly understood. And yet I feel a > little bit disappointed. Of course it is interesting, > but, how should I say ? I have the impression that, > despite of the title, the author beholds a tendency to > think "syntactically" (and more, English-oriented > syntactically) rather than "semantically".
IMO Jackendoff thinks both too syntactically and not syntactically enough. I think the relationship between syntax and 'ideation' is exactly analogous to the relationship between phonology (which is part of langue) and phonetics (which is not). In the same way that phonological representations are interpreted phonetically by very simple rules of phonetic realization, and so must resemble phonetic forms, so syntactic representations are interpreted ideationally by very simple rules of 'ideational realization', so must resemble ideational forms. Jackendoff by contrast has very superficial syntactic structures bearing little resemblance to ideational forms and these syntactic structures map by complicated langue-internal correspondence rules to supposedly universal conceptual structures that operate according to a quasi-grammar & quasi-lexicon of their own that Jackendoff, for no very apparent reason, seems to believe is discoverable through standard methodologies of theoretical linguistics.
> Maybe I'm > wrong, but for ex, when I came across Chapter 7 (The > Action Tier and the Analysis of Causation), I felt > very perplex about an example he gives (in fact he > seems to refer to Anderson, but he doesn't add any > comment, except that "unfortunately", Anderson used > the term "Theme" rather than "Patient"): > > The point (as I understand it) is about comparing: > - Bill loaded the books onto the truck > - Bill smeared paint onto the wall. > So we try to find out what is the Patient in both > cases, by reformulating these sentences like: > > - What Bill did to the books was load them on the > truck > - What Bill did to the truck was load it with books > - What Bill did to the paint was smear it on the wall > - What Bill did to the wall was smear paint on it [and > more] > > and the conclusion is apparently: "The thematic > relations in each case are the same: the books go onto > the truck, the paint goes onto the wall. The change is > in which entity is viewed as most directly "affected" > by Bill's action, and the direct object has a stronger > claim on the role in either case [...]
He is comparing them with Bill loaded the truck with the books. Bill smeared the wall with paint. these are "the two syntactic frames of _load_ and _smear_" (p129). This is not at all apparent from what he actually says on pp129-130, unless you the reader already happen to know that the alternation is a long-standing topos in lexical semantics. The thematic relations (in J's terminology) are the Theme/Goal relations. It is widely held that in "smear the wall with paint" and "load the truck with the books" the wall/truck is more completely affected than in the other version.
> I wonder about these "thematic relations being the > same". To me, it's very different, conceptually > speaking. In one case Bills move a physical thing from > one place to another (and lays it according to the > laws of gravity), in the other case Bill changes the > external appearance of a (vertical) part of something, > by means of applying a (liquid) substance on it. What > has it to do together ? The answer is clear: the > syntax is the same. Huh, I thought we were talking > about semantics ? Does really the single word "onto" > imply an equivalent meaning in both cases ? To me, > "onto" just implies: direction + contact. This is true > in both cases, and yet the situations seem very > different.
The situations are very different, but language is a tool that must be able to serve for every occasion. Therefore language cannot encode every conceivable nuance of meaning and must instead paint the world in broader strokes, with a degree of abstraction such that the two situations are very similar. "Semantics" is (in linguistics) a *linguistic* notion: semantics cares about the nature of smearing, paint and walls only if the rules of the language are sensitive to it, and we find that the grammar of English has nothing to say about smearing, paint or walls, but does have something to say about Themes and Goals.
> Besides, reformulations like "What Bill did to the > paint was smear it on the wall" makes little sense to > me. I could have said "Bill smeared the wall with > paint". Where is the Patient ? Is there any Patient at > all ? Are they two of them ?
Jackendoff pretty much defines the linguistic notion Patient as that which can be expressed by X in the "do to X" construction.
> The author refers to "the > syntactic frames of verbs like 'load' and 'smear'". > This is, once more, about syntax, and of course, about > some special syntax, the English one. To me, if we > want to talk about semantics, we should do it the > opposite way: starting from the possible concepts, > whatever the natlang, and then come to the > implementations in languages.
It is extraordinarily difficult to start from the possible concepts, since we have no credible methodology for establishing what they are, except in the vaguest terms. Also, as I have said above, for many linguists 'semantics' means not our conceptual inventory but only what is linguistically encodable.
> For ex: let's take a > census of all the possible spatial concepts (with the > help of drawings), then let's see how English (and > other languages) manage them. Seems to me that > Jackendoff is thinking all the time "syntactically".
Yes, because syntax is the interface between language and concepts.
> He also seems to take for granted, without any further > inquiry, that "major ontological categories", or > "conceptual parts of speech", are "Thing, Event, > State, Action, Place, Path, Property, and Amount". To > me it is clear that Human Being, for ex, can hardly be > described just as a "Thing", even having special > properties. To me Human Being IS a major ontological > category, even if it shares some properties with > Animals or Artefacts or Natural Objects. (And also, > where is Modality for ex ?) He also thinks in terms of > "arguments", for ex: "to run" implies a Path (if I > remember well), and if the path (the direction for ex) > is not expressed, than it is understood from the > context (I can't find the exact quotation). In other > terms, if you run, than you run toward some place, or > from some place, or whatever, but you can't just > "run". I disagree: "What have you been doing this > morning ? - Running" (meaning: having physical > exercise by the action of running). Maybe I ran on a > training outfit, not moving one step forwards ! In "to > run", there are clearly two concepts: 1/ to go in a > fast way; 2/ to move one's body in a special way. You > can focus on either concept, or on the combined one. > Only if you focus on concept 1, or on concept 1+2, you > may say that there is a Path implied. (Maybe this is > even more evident with "to swim").
Certainly one can raise many such objections to J's work, not only to specific details, but also to the flimsiness of its foundations -- it is very unclear how one could demonstrate that any given detail is wrong (which is perhaps why J has many admirers but few followers).
> Maybe I got it all wrong and it will become clearer in > the following chapters, but until now, I feel a little > disappointed.
I think you've got it approximately right, but your problem is that your interest is in not semantics but cognition. But there are scholars who work on how we conceptualize the world around us -- a lot of so called Cognitive 'Linguistics' is about this, & I would particularly recommend to you the work of Len Talmy as likely to appeal. (When I first arrived at university to study linguistics my primary interest was in cognitive poetics, which simply didn't exist at the time, but then I went through very much the process you seem to be going through: an interest in what I then thought of as 'semantics' led me, via Jackendoff, after a long & frustrating search in the dark, to the then brand-new work of Langacker. I began my doctoral work as a fully paid-up cognitive linguist who viewed formal syntax with nausea and disdain, and spent the following too-many years becoming ever more disenchanted with cognitivism and mutating into a formal syntactian.) --And.


Philippe Caquant <herodote92@...>